Microsoft CEO Satya Nadella says we should treat all AI models as potentially compromised, a security-first stance with major implications for trust, authenticity, and how enterprises deploy generative AI systems.
Nikon disqualified a Small World in Motion competition winner after discovering the submission used generative AI, reigniting debate over authenticity, disclosure, and how contests detect synthetic media.
A North Carolina musician admitted to using hundreds of thousands of AI-generated songs and thousands of bot accounts to fraudulently inflate streams, netting over $10 million in royalties in one of the first cases of its kind.
A new lightweight framework promises rigorous deepfake detection on modest hardware, lowering the barrier to entry for researchers, smaller organizations, and edge deployments battling synthetic media.
The original transformer had both an encoder and a decoder. GPT threw away the encoder entirely. Here's why decoder-only architecture became the backbone of modern generative AI — and what it means for the models powering today's synthetic media.
New research reveals that LLM-based fact-checkers may inflate their misinformation detection scores through data leakage—exploiting hindsight knowledge rather than genuine reasoning. The findings have major implications for digital authenticity benchmarks.
Alibaba's Qwen has grown from a modest 7B-parameter model into a sprawling family reaching 2.4 trillion parameters, spanning language, vision, audio, and multimodal generation that increasingly shapes the open-weight AI landscape.
New research probes how robust multimodal fact-checking systems are when claims are rewritten by LLMs. The findings reveal fragility in verification pipelines that could let manipulated or synthetic content slip past automated authenticity checks.
Generative Adversarial Networks power much of today's synthetic media, from deepfakes to AI-generated faces. Here's a technical breakdown of how GANs pit two neural networks against each other to learn the art of creating realistic new data.
Suno, the popular AI music generator, is rolling out a beta speech feature that turns text into spoken audio, expanding from song creation into full synthetic voice generation and raising fresh questions about audio authenticity.
A new research paper proposes manifold-constrained initial noise optimization, a lightweight method to align diffusion-based generative models with human preferences and reward signals without costly fine-tuning.
A new arXiv paper argues that defining the null hypothesis is the hardest part of detecting memorization in generative models, proposing exact statistical tests to rigorously determine when a model is reproducing its training data.